Alpha-Mangostin Reverses Multidrug Resistance by Attenuating the Function of the Multidrug Resistance-Linked ABCG2 Transporter

Abstract

The ATP-binding cassette (ABC) drug transporter ABCG2 can actively efflux a wide variety of chemotherapeutic agents out of cancer cells and subsequently reduce the intracellular accumulation of these drugs. Therefore, the overexpression of ABCG2 often contributes to the development of multidrug resistance (MDR) in cancer cells, which is one of the major obstacles to successful cancer chemotherapy. Moreover, ABCG2 is highly expressed in various tissues including the intestine and blood-brain barrier (BBB), limiting the absorption and bioavailability of many therapeutic agents. For decades, the task of developing a highly effective synthetic inhibitor of ABCG2 has been hindered mostly by the intrinsic toxicity, the lack of specificity, and complex pharmacokinetics. Alternatively, considering the wide range of diversity and relatively nontoxic nature of natural products, developing potential modulators of ABCG2 from natural sources is particularly valuable. α-Mangostin is a natural xanthone derived from the pericarps of mangosteen (Garcinia mangostana L.) with various pharmacological purposes, including suppressing angiogenesis and inducing cancer cell growth arrest. In this study, we demonstrated that at nontoxic concentrations, α-mangostin effectively and selectively inhibits ABCG2-mediated drug transport and reverses MDR in ABCG2-overexpressing MDR cancer cells. Direct interactions between α-mangostin and the ABCG2 drug-binding site(s) were confirmed by stimulation of ATPase activity and by inhibition of photolabeling of the substrate-binding site(s) of ABCG2 with [¹²⁵I]iodoarylazidoprazosin. In summary, our findings show that α-mangostin has great potential to be further developed into a promising modulator of ABCG2 for reversing MDR and for its use in combination therapy for patients with MDR tumors.

Graphical Abstract

INTRODUCTION

The ATP-binding cassette (ABC) drug transporters can utilize ATP hydrolysis to actively transport a wide variety of therapeutic agents out of cancer cells, rendering chemotherapy ineffective and increase the risk of cancer relapse and death.¹ Therefore, the development of multidrug resistance (MDR) in cancer is often associated with overexpression of an ABC transporter, most commonly ABCB1(P-glycoprotein/MDR1) or ABCG2 (BCRP; MXR).^2,3 Human ABCG2 is one of the latest discovered major ABC drug transporters to be involved in MDR phenotype.^4,5 ABCG2 is capable of effluxing and conferring resistance to most conventional anticancer agents, and some molecularly targeted drugs such as tyrosine kinase inhibitors.^3,6–8 Consequently, overexpression of ABCG2 is often linked to MDR in patients with advanced nonsmall cell lung cancer or leukemia.^1,6,8,9 Moreover, ABCG2 is localized at many important biological barriers, such as the luminal membrane of brain capillaries and the blood-brain barrier (BBB), protecting cells or tissues from xenobiotics and harmful chemotherapeutics, thus affecting the overall drug bioavailability, distribution, metabolism, and elimination.^3,10,11 Therefore, modulating the function or expression of ABCG2 has great clinical significance.

One of the most effective ways to overcome MDR in ABCG2-overexpressing cancer cells at present is by directly inhibiting the function of ABCG2.¹² The idea is to introduce a selective modulator at a nontoxic concentration to transiently inhibit the transport function of ABCG2 in a competitive manner, elevating the accumulation of conventional anticancer drug in MDR cancer cells.³ Unfortunately, many failed attempts using immunosuppressants, channel blockers and synthetic chemicals to modulate the function of ABCG2 were associated with the lack of specificity, high intrinsic toxicity, and adverse drug–drug interactions.^3,13,14 As a result, none of the MDR modulators from the first three generations has been approved to treat patients with MDR tumors,¹³ and the discovery of modulators originating from natural sources is underway.¹⁵ The advantage of utilizing compounds originating from natural sources is that they are generally low in toxicity, better tolerated in human body, and offer the widest range of diversity and novel chemical scaffolds.¹⁵ Therefore, it is not surprising that natural products are being used by most cancer patients in combination with anticancer drugs for the potential benefits in chemoprevention and delaying cancer progression.¹⁶

α-Mangostin (see Figure 1 for structure) is one of the most studied antioxidant phytochemicals^17–21 and the most abundant xanthone derived from the pericarps of mangosteen (Garcinia mangostana L.),²² a popular fruit in Southeast Asia. Many biological properties of α-mangostin have been reported, including cardioprotective, neuroprotective, antidiabetic, anti-inflammatory, antimicrobial, and antiprotozoal activities.^23,24 In addition, the antitumor activity of α-mangostin has been extensively studied in in vitro and in vivo systems.²⁵ α-Mangostin has been shown to suppress tumor progression through induction of cell cycle arrest,^21,26,27 apoptosis,^26,28,29 and autophagy,³⁰ as well as inhibition of angiogenesis,^31,32 invasion, and metastasis.^27,31–34 Moreover, α-mangostin has been reported to be able to selectively target cancer cells without having cytotoxic effect to normal cells^35,36 by acting synergistically with numerous conventional chemotherapeutic agents, including doxorubicin³⁷ and 5-FU.³⁸

Figure 1.

Chemical structure of α-mangostin.

In the present study, we demonstrated for the first time that α-mangostin interacts directly with MDR-linked ABC transporter ABCG2 and inhibits its transport function without affecting the protein expression level of ABCG2. More importantly, α-mangostin significantly resensitizes ABCG2-overexpressing MDR cancer cells to chemotherapeutic drugs.

EXPERIMENTAL SECTION

Chemicals.

Dulbecco’s Modified Eagle’s medium (DMEM), RPMI medium, fetal calf serum (FCS), phosphate-buffered saline (PBS), trypsin-EDTA, penicillin, and streptomycin were purchased from Gibco, Invitrogen (CA, USA). Tariquidar was obtained from MedKoo Biosciences (Morrisville, NC). [¹²⁵I]-Iodoarylazidoprazosin (IAAP) (2200 Ci/mmol) was from PerkinElmer Life Sciences. All other chemicals were purchased from Sigma-Aldrich Company (St. Louis, MO, USA).

Cell Lines and Culture Conditions.

pcDNA3.1-HEK293, ABCB1-transected MDR19-HEK293, ABCG2-transfected R482-HEK293, KB-3–1, KB-V-1, MCF-7, and MCF7-FLV1000 were cultured in DMEM, whereas S1 and S1-M1–80 cells were cultured in RPMI-1640 (Gibco, Invitrogen). All cell lines were maintained at 37 °C in culture media containing 10% FCS, 2 mM L-glutamine, and 100 units of penicillin/streptomycin/mL in 5% CO₂ humidified air. HEK293 and HEK293 transfected lines were maintained in 2 mg/mL G418,³⁹ KB-V-1 cells were maintained in media containing 1 mg/mL vinblastine,⁴⁰ and MCF7-FLV1000 cells were cultured in the presence of 1 μg/mL flavopiridol,^41,42 whereas S1-M1–80 cells were cultured in 80 μM of mitoxantrone.³⁹ Cells were placed in drug-free medium 7 days prior to assay.

Fluorescent Drug Accumulation Assay.

A FACSort flow cytometer equipped with Cell Quest software (Becton-Dickinson) was used to monitor the intracellular accumulation of fluorescent substrates as described previously.⁴³ Briefly, cells were harvested and resuspended in Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 5% FCS. Calcein-AM or pheophorbide A (PhA) was added to 3 × 10⁵ cells in 4 mL of IMDM in the presence or absence of α-mangostin or respective reference inhibitors to determine ABCB1-mediated calcein-AM efflux or ABCG2-mediated PhA efflux according to the method described by Gribar et al.⁴⁴

Cytotoxicity Assay.

Cell Counting Kit-8 (CCK) assay was used to determine the cytotoxicity of drugs in HEK293 and HEK293 transfected with ABCB1 (MDR19-HEK293) or ABCG2 (R482-HEK293), whereas MTT assay was used to determine the cytotoxicity of drugs in human cancer cell lines according to the method described by Ishiyama et al.⁴⁵ For the reversal of cytotoxicity assays, a nontoxic concentration of α-mangostin or reference inhibitor of ABCB1 or ABCG2 was added into the cytotoxicity assay, and the extent of reversal was determined based on the calculated fold-reversal (FR) values.

ATPase Assay and Photoaffinity Labeling of ABCG2 with [¹²⁵I]IAAP.

The effect of α-mangostin on vanadate (Vi)-sensitive ATPase activity of ABCG2 was determined using crude membranes isolated from High-Five cells based on the end point P_i assay as described previously.⁴⁶ Photoaffinity labeling assays were performed using membranes (50–75 μg protein/mL) prepared from ABCG2-overexpressing MCF7-FLV1000 cells as described previously.⁴⁶ Membranes were first incubated with α-mangostin for 10 min at room temperature in 50 mM Tris-HCl, 150 mM NaCl, pH 7.5, before 3–6 nmol/L [¹²⁵I]IAAP (2200 Ci/mmol) was added. The samples were then photo-cross-linked with UV light (366 nm) and processed as described previously.³⁹

Immunoblotting.

Antibodies C219 (1:1000), BXP-21 (1:500) and α-tubulin (1:2000) were used in Western blot immunoassay to detect ABCB1, ABCG2 and tubulin, respectively. The horseradish peroxidase-conjugated goat antimouse IgG (1:10,000) was used as the secondary antibody. Signals were detected as described previously.³⁹

Molecular Modeling of ABCG2.

For structure prediction, the three-dimensional structure of ABCG2 was predicted using SWISS-MODEL, an automated protein homology-modeling server. The amino acid sequence of the protein was submitted to SWISS-MODEL server and templates were searched with BLAST and HHBlits against SWISS-MODEL template library. For each identified template, the template’s quality was predicted from features of the target-template alignment. The templates with the highest quality were then selected for model building.^47–49 Models were built based on the target-template alignment using ProMod3. For ligand preparation and docking, the energy was minimized for both protein and ligand using Acclerys Discovery Studio 4.0 and docking was also performed by the same software.

Statistical Analysis.

Experimental data and IC₅₀ values were obtained from at least three independent experiments, and presented as mean ± standard error of the mean (SEM) and mean ± standard deviation (SD), respectively. Two-sided Student’s t test was used to determine the differences between any mean values, and results were considered statistically significant at P < 0.05.

RESULTS

α-Mangostin Inhibits ABCG2-Mediated Substrate Transport.

First, the effect of α-mangostin on the transport function of ABCB1and ABCG2 was determined as described in Experimental Section. As shown in Figure 2A, ABCB1-mediated efflux of fluorescent substrate calcein was unaffected by 5 μM α-mangostin in ABCB1-transfected MDR19-HEK293. In contrast, ABCG2-mediated efflux of fluorescent PhA, a known fluorescent substrate of ABCG2,⁴³ was strongly inhibited by α-mangostin in ABCG2-transfected R482-HEK293 (Figure 2B). Similarly, α-mangostin inhibited ABCG2-mediated PhA transport in ABCG2-overexpressing human S1-M1–80 colon cancer (Figure 2C) and human MCF7-FLV1000 breast cancer (Figure 2D) cell lines. As positive controls, 3 μM of tariquidar and 1 μM of Ko143 (dotted lines) were used in the assay as specific reference inhibitors for ABCB1 and ABCG2, respectively. Of note, α-mangostin had no significant effect on the accumulation of fluorescent probes in any of the drug sensitive parental cells tested (Figure 2A–D, left panels).

Figure 2.

Effect of α-mangostin on the fluorescent substrate transport mediated by ABCB1 or ABCG2. The accumulation of calcein-AM in parental HEK293 (A, left panel) and ABCB1-transfected MDR19-HEK293 cells (A, right panel), as well as the accumulation of pheophorbide A (PhA) in parental HEK293 (B, left panel) and ABCG2-transfected R482-HEK293 cells (B, right panel), in S1 (C, left panel) and ABCG2-overexpressing S1-M1–80 (C, right panel), or in MCF-7 (D, left panel) and ABCG2-overexpressing MCF7-FLV1000 cancer cells (D, right panel), were measured in the absence (solid lines) or presence of 5 μM α-mangostin (shaded, solid lines) or a reference inhibitor (dotted lines) of ABCB1 (3 μM tariquidar) or ABCG2 (1 μM Ko143). Cells were analyzed immediately by flow cytometry as described in Experimental Section. Representative histograms and immunoblots of ABCB1 (A, inset) or ABCG2 (B–D, inset) in total cell lysate protein (10 μg) from pcDNA-HEK293, MDR19-HEK293, R482-HEK293, S1, S1-M1–80, MCF7, and MCF7-FLV1000 cells are shown.

α-Mangostin Reverses MDR in Cells Overexpressing ABCG2.

Knowing that α-mangostin inhibits the function of ABCG2, we next examined the reversal effect of α-mangostin on ABCG2-mediated MDR in ABCG2-overexpressing cells. Without affecting the proliferation of drug sensitive cells (Figure 3, left panels), α-mangostin restored the sensitivity of ABCG2-transfected R482-HEK293 cells to topotecan (Figure 3A, right panel) and mitoxantrone (Figure 3B, right panel), two well-established anticancer drug substrates of ABCG2, in a concentration-dependent manner. At a highest tested concentration of 2 μM, α-mangostin resensitized R482-HEK293 cells to topotecan and mitoxantrone by approximately 25-fold and 9-fold, respectively. Similarly, multiple nontoxic concentrations of α-mangostin were also tested in ABCG2-overexpressing S1-M1–80 cancer cells, and the reversal effect of α-mangostin on ABCG2-mediated MDR is summarized in Table 1. In contrast, α-mangostin was unable to resensitize ABCB1-overexpressing MDR19-HEK293 cells and KB-V-1 cancer cells to ABCB1 substrates⁵⁰ doxorubicin (Figure 3C) and colchicine (Figure 3D). The fold-reversal (FR) value indicates the degree of cellular chemosensitivity restored by a particular reversing agent, which was calculated as described previously.⁵¹ Tariquidar and Ko143 were used as positive controls to reverse MDR conferred by ABCB1 or ABCG2, respectively.⁵² Our results indicate that α-mangostin is selective for ABCG2 relative to ABCB1.

Figure 3.

Effect of α-mangostin on reversing multidrug resistance mediated by ABCG2 or ABCB1. Cytotoxicity of topotecan (A) or mitoxantrone (B) in HEK293 cells (left panels) and ABCG2-transfected R482-HEK293 cells (right panels), as well as doxorubicin (C) or colchicine (D) in HEK293 cells (left panels) and ABCB1-transfected MDR19-HEK293 cells (right panels) was determined in the absence (open circles) or presence of α-mangostin at 0.5 μM (filled circles), 1.0 μM (open squares), 2.0 μM (filled squares), or 1 μM of reference inhibitor Ko143 or tariquidar (open triangles). Points: means from at least three independent experiments; bars: SEM.

Table 1.

Effect of α-Mangostin on ABCG2- and ABCB1-Mediated Multidrug Resistance in Transfected and Drug-Selected Cell Lines^a

treatment	concentration (μM)	mean IC50b ± SD and (FRc)
		pcDNA- HEK293 [nM]	R482-HEKS293 [nM]
topotecan		17.90 ± 4.21 (1.0)	627.68 ± 135.29 (1.0)
+α-mangostin	0.5	18.28 ± 4.46 (1.0)	252.47 ± 39.59** (2.5)
+α -mangostin	1	19.18 ± 5.03 (0.9)	73.15 ± 8.21** (8.6)
+α -mangostin	2	18.84 ± 4.47 (1.0)	25.21 ± 5.56** (24.9)
+Ko143	1	19.06 ± 4.93	24.43 ± 7.46** (25.7)
mitoxantrone		1.93 ± 0.41 (1.0)	136.14 ± 27.99 (1.0)
+α -mangostin	0.5	2.05 ± 0.44 (0.9)	84.53 ± 15.80* (1.6)
+α -mangostin	1	2.20 ± 0.45 (0.9)	35.47 ± 4.73** (3.8)
+α -mangostin	2	2.52 ± 0.48 (0.8)	15.62 ± 2.60** (8.7)
+Ko143	1	1.72 ± 0.32 (1.1)	8.09 ± 1.88** (16.8)
		S1 [nM]	S1-M1–80 [μM]
topotecan		29.18 ± 7.29 (1.0)	7.60 ± 2.09 (1.0)
+α -mangostin	0.5	27.69 ± 6.05 (1.1)	2.17 ± 0.41* (3.5)
+α -mangostin	1	31.93 ± 6.99 (0.9)	0.77 ± 0.11** (9.9)
+α -mangostin	2	25.75 ± 6.00 (1.1)	0.53 ± 0.14** (14.3)
+Ko143	1	31.83 ± 7.82 (0.9)	0.40 ± 0.11** (19.0)
mitoxantrone		8.87 ± 1.91 (1.0)	89.98 ± 16.19 (1.0)
+ α -mangostin	0.5	11.15 ± 2.06 (0.8)	55.58 ± 13.69* (1.6)
+ α -mangostin	1	10.67 ± 1.94 (0.8)	23.86 ± 4.36** (3.8)
+α -mangostin	2	8.45 ± 1.98 (1.0)	9.11 ± 2.04** (9.9)
+Ko143	1	7.90 ± 1.84 (1.1)	1.04 ± 0.16*** (86.5)
		pcDNA-HEK293 [nM]	MDR19-HEKS293 [nM]
doxorubicin		11.84 ± 2.57 (1.0)	179.08 ± 48.63 (1.0)
+α -mangostin	2	10.99 ± 2.54 (1.1)	226.67 ± 56.13 (0.8)
+tariquidar	1	6.34 ± 1.49* (1.9)	20.43 ± 5.13* (8.8)
colchicine		16.22 ± 5.83 (1.0)	317.87 ± 89.54 (1.0)
+α -mangostin	2	13.67 ± 4.18 (1.2)	228.31 ± 59.00 (1.4)
+tariquidar	1	9.28 ± 3.79 (1.7)	14.15 ± 4.47** (22.5)
		KB-3–1 [nM]	KB-V-1 [μM]
doxorubicin		30.60 ± 7.95 (1.0)	1.91 ± 0.37 (1.0)
+α -mangostin	2	16.90 ± 5.49 (1.8)	1.24 ± 0.33 (1.5)
+tariquidar	1	26.13 ± 6.07 (1.2)	42.44 ± 10.48 [nM]*** (45.0)
colchicine		31.67 ± 10.53 (1.0)	1.64 ± 0.15 (1.0)
+α -mangostin	2	29.88 ± 9.35 (1.1)	1.54 ± 0.19 (1.1)
+tariquidar	1	30.89 ± 9.95 (1.0)	44.20 ± 14.59 [nM]*** (37.1)

^aAbbreviation: FR, fold-reversal.

^bIC₅₀ values are mean ± SD calculated from dose-response curves obtained from three independent experiments using cytotoxicity assay as described in Experimental Section.

^cFR values were obtained by dividing IC₅₀ values of cells treated with a given anticancer drug in the absence of a modulator by IC₅₀ values of cells treated with the same anticancer drug in the presence of a modulator.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

ABCG2 Does Not Confer Resistance to α-Mangostin in Cancer Cells.

Knowing that α-mangostin interacts with ABCG2, we next examined whether overexpression of ABCG2 in cancer cells confers significant resistance to α-mangostin. The cytotoxicity of α-mangostin was determined in several drug-sensitive and ABCG2-overexpressing MDR cancer cell lines, and the calculated IC₅₀ values are summarized in Table 2. The resistance factor (RF) value was calculated by dividing the IC₅₀ value of α-mangostin in ABCG2-overexpressing MDR subline by the IC₅₀ value of α-mangostin in the respective parental line, representing the extent of cellular resistance to α-mangostin caused by the overexpression of ABCG2. We did not observe significant differences in α-mangostin IC₅₀ values in ABCG2-overexpressing cancer cell lines in respective to the drug-sensitive cancer cell lines. In addition, the cytotoxicity of α-mangostin was also determined in HEK293 cells and ABCG2-transfected R482-HEK293 cells to confirm our findings. We found that both cell lines were equally sensitive to α-mangostin (Table 2).

Table 2.

Cytotoxicity of α-Mangostin in Drug-Sensitivity and ABCG2-Overexpressing Multidrug Resistant Cells

cell line	cancer origin	transporter expressed	IC50 (nM)a	RFb
S1	colon		10.78 ± 4.48	1.0
S1-M1–80	colon	ABCG2	15.43 ± 5.63	1.4
H460	lung		11.58 ± 5.14	1.0
H460-MX20	lung	ABCG2	10.42 ± 4.39	0.9
A549	lung		13.04 ± 4.54	1.0
A549-Bec150	lung	ABCG2	10.84 ± 4.18	0.8
MCF7	breast		12.58 ± 6.97	1.0
MCF7-FLV1000	breast	ABCG2	13.18 ± 4.09	1.0
MCF7-AdVp3000	breast	ABCG2	17.89 ± 4.06	1.4
pcDNA-HEK293			17.31 ± 6.94	1.0
R482-HEK293		ABCG2	20.37 ± 7.64	1.2

^aIC₅₀ values are mean ± SD calculated from dose-response curves obtained from three independent experiments using cytotoxicity assay as described in Experimental Section.

^bRF values were calculated by dividing IC₅₀ values of ABCG2-overexpressing MDR cells by IC₅₀ values of drug-sensitive parental cells.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

α-Mangostin Does Not Affect the Protein Expression of ABCG2.

In addition to direct inhibition of ABCG2 transport function, drug-induced transient down-regulation of ABCG2 can potentially increase the drug sensitivity of ABCG2-overexpressing cancer cells to chemotherapeutics.^53,54 Therefore, we examined protein expression of ABCG2 after exposing human S1-M1–80 colon cancer cells to increasing concentrations of α-mangostin (0–3 μM) for 72 h and processed by immunoblotting as described in Experimental Section. As shown in Figure 4, α-mangostin does not significantly alter the protein expression of ABCG2 in S1-M1–80 cancer cells. Our results indicate that α-mangostin reverses ABCG2-mediated MDR by inhibiting the function of ABCG2, and not by down regulation of ABCG2.

Figure 4.

Effect of α-mangostin on ABCG2 protein expression in human S1-M1–80 colon cancer cells. (A) Immunoblot detection of human ABCG2 and (B) quantification of total lysate protein (10 μg) from drug-resistant S1-M1–80 cells treated with increasing concentrations (0–3 μM) of α-mangostin (MG) for 72 h as described previously.⁷⁵ α-Tubulin was used as loading control. Values are presented as mean ± SEM calculated from three independent experiments.

Effect of α-Mangostin on [¹²⁵I]IAAP Photoaffinity Labeling and ATPase Activity of ABCG2.

To study the interaction of α-mangostin with the substrate-binding sites of ABCG2, we examined the effect of α-mangostin on photoaffinity labeling of ABCG2 with [¹²⁵I]IAAP and vanadate (Vi)-sensitive ATPase activity of ABCG2. We found that 10 μM of α-mangostin inhibited the incorporation of [¹²⁵I]IAAP into ABCG2 by approximately 60% (Figure 5A), indicating direct binding of α-mangostin to the substrate-binding sites of ABCG2. Moreover, considering that substrate transport mediated by ABC transporter is coupled to ATP hydrolysis,^55,56 the effect of α-mangostin on Vi-sensitive ATPase activity of ABCG2 in membranes isolated from cells expressing ABCG2 was examined in order to gain insight into the interaction between α-mangostin and ABCG2. As shown in Figure 5B, α-mangostin stimulated ABCG2 ATPase activity in a concentration-dependent manner, with approximately 2-fold maximum stimulation at 1 μM and with the concentration of 86 ± 16 nM (Figure 5B, inset) for 50% stimulation.

Figure 5.

Effect of α-mangostin on photoaffinity labeling of ABCG2 with [¹²⁵I]IAAP and vanadate (Vi)-sensitive ABCG2 ATPase activity. (A) Membrane protein (250–500 μg protein/mL) from MCF7-FLV1000 cells expressing ABCG2 was incubated in the presence or absence of 10 μM of α-mangostin at 21 °C in 50 μmol/L Tris-HCL (pH 7.5), 150 mM NaCl. [¹²⁵I]IAAP (2,200 Ci/mmol) at 3 to 6 nmol/L was added to the samples and incubated for 5 min under subdued light, then illuminated with an UV lamp (365 nm) for 10 min at room temperature. The labeled ABCG2 was processed and visualized as described previously.³⁹ Representative autoradiogram (A, inset) and values represent means and standard deviations calculated from three independent experiments. (B) Membrane protein (50–100 μg protein/mL) from High-Five insect cells expressing ABCG2 was incubated at 37 °C with 0–10 μM of α-mangostin and 0–1 μM of α-mangostin (B, inset) in the presence or absence of vanadate. The ABCG2 ATPase activity was measured as described previously.³⁹ Points: mean from at least three independent experiments; bar: SEM.

α-Mangostin Docking Analysis with Molecular Modeling of ABCG2.

To better understand the binding of α-mangostin with ABCG2 transporter protein, docking study was performed using a homology model generated from currently available ABCG5 crystal structure as template.^57,58 The potential binding site for α-mangostin lies within the transmembrane domain (TMD) of ABCG2 and the binding was mostly stabilized by hydrophobic interactions with surrounding hydrophobic residues Pro⁴⁸⁵, Val⁴⁴², Ala⁴⁴⁴, Val⁴⁴⁵, Val⁴⁰¹, Ile³⁹⁹, Ile⁴⁰⁰, and Ala³⁹⁷. A hydrogen bond was formed between a phenolic hydroxyl group and the carbonyl moiety of Ala³⁹⁷ (Figure 6).

Figure 6.

Binding mode of α-mangostin in the homology model of ABCG2 by using Acclerys Discovery Studio 4.0 software as described in Experimental Section. α-Mangostin is shown as ball and stick model with the atoms colored as carbon-dark gray, hydrogen-light gray, nitrogen-blue, and oxygen-red. The same color scheme is used for interacting amino acid residues. Dotted green line indicates proposed hydrogen bond.

DISCUSSION

ABCG2-mediated multidrug resistance to chemotherapeutic agents remains a substantial challenge to the effective treatment of cancer.⁵⁹ Instead of developing novel synthetic compounds or using clinically active drugs at higher than intended concentrations as reversing agents, we and others have explored utilizing bioactive compounds originated from natural sources to overcome ABCG2-mediated MDR in cancer cells.¹⁵ Although many natural product modulators have been identified, most of them lack the combination of potency and selectivity.^13,15 Moreover, some of these modulators are toxic in nature, and therefore unsuitable for clinical use.¹⁵ For instance, fumitremorgin C (FTC) has one of the highest in vitro activity against ABCG2,⁶⁰ however, the neurotoxic nature of FTC prevented it from clinical applications.⁶¹ The natural xanthone α-mangostin is one of the well-characterized bioactive natural compounds, known for its antiperoxidative, anti-inflammation and anticancer properties.^21,25,62,63

In the present study, we studied the potential modulatory effect of α-mangostin on ABCG2-mediated MDR in cancer cells. Several short-term drug accumulation assays were performed to evaluate the specificity and inhibitory activity of α-mangostin against drug efflux mediated by MDR-linked ABC drug transporters ABCB1 and ABCG2. We noticed that ABCG2-mediated drug efflux was significantly inhibited by α-mangostin. In contrast, the effect of α-mangostin on ABCB1-mediated drug efflux was extremely weak (Figure 2). Our results indicate that α-mangostin selectively inhibits the function of ABCG2. Next, we explored whether inhibition of ABCG2 transport function by α-mangostin could lead to resensitization of ABCG2-overexpressing cells to chemotherapeutic drugs in the same manner as other natural product modulators, such as curcumin,⁶⁴ resveratrol,⁶⁵ and plumbagin.⁶⁶ We found that at nontoxic concentrations, α-mangostin substantially resensitized ABCG2-overexpressing cells to topotecan and mitoxantrone. Interestingly, α-mangostin appears to be more effective in restoring the sensitivity of ABCG2-overexpressing cells to topotecan than mitoxantrone (Table 1). Of note, in spite of the weak inhibition ABCB1-mediated efflux by α-mangostin, there was still the possibility of α-mangostin reversing MDR in ABCB1-overexpressing cells. Therefore, we also examined the effect of α-mangostin on ABCB1-mediated resistance to doxorubicin and colchicine. As expected, α-mangostin was unable to restore chemosensitivity in ABCB1-overexpressing cells (Table 1). Considering the overlapping substrate specificity of ABCB1 and ABCG2,^3,67 it is not surprising that although many natural product modulators showed promising results in restoring the chemosensitivity of MDR cancer cells, most were not transporter-specific.¹⁵ For instance, curcumin is known to resensitize ABCG2-overexpressing cancer cells to chemotherapy, but it also resensitizes ABCB1-overexpressing cells in a same manner.⁶⁸ An alternative way to resensitize ABC transporter-overexpressing MDR cancer cells is to reduce transiently the expression of that particular ABC drug transporter.^54,69,70 We discovered that α-mangostin did not alter the protein expression of ABCG2 in ABCG2-overexpressing MDR cancer cells over a period of 72 h (Figure 4), supporting the notion that α-mangostin resensitizes ABCG2-overexpressing MDR cancer cells to anticancer drugs by inhibiting the function of ABCG2. Next, we examined whether ABCG2 is likely to confer resistance to α-mangostin by comparing the intrinsic cytotoxicity of α-mangostin in multiple drug-sensitive cell lines and respective ABCG2-overexpressing MDR sublines. We discovered that ABCG2-overexpressing human colon, lung, and breast cancer cells, as well as ABCG2-transfected cells were all equally sensitive to α-mangostin as their respective parental cells (Table 2). Again, these results show that α-mangostin behaves as a high-affinity competitive modulator of ABCG2 and not a transport substrate of ABCG2.

To confirm the direct interaction between α-mangostin and substrate-binding site(s) of ABCG2, photoaffinity labeling and ATPase assay of ABCG2 experiments were performed. [¹²⁵I]IAAP is a transport substrate that binds directly to the substrate binding sites of ABCG2,⁷¹ and any substrate or inhibitor that binds to the same site will inhibit the photolabeling.^39,72 Our photoaffinity labeling results demonstrated direct and competitive binding of α-mangostin to the substrate-binding pocket(s) of ABCG2. Moreover, substrate transport by an ABC transporter is known to be coupled with the stimulation of ATPase activity,⁷³ where rapid stimulation of ATP hydrolysis is associated with the presence of a substrate, while inhibition of ATP hydrolysis is associated with the presence of an inhibitor or substrate with a much lower transport rate.⁷⁴ α-Mangostin produced a sharp stimulation at low concentrations (0.5–1 μM) and decreased stimulation of the ABCG2 ATPase activity at higher concentrations (2–10 μM), suggesting that α-mangostin exhibits a high affinity for ABCG2. These results demonstrated that α-mangostin binds to the substrate-binding pocket(s) of ABCG2 and attenuates the transport function of ABCG2 in a competitive manner.

Since high-resolution crystal structure of human ABCG2 is still unavailable, an ABCG2 homology model was used instead to better understand the binding of α-mangostin with ABCG2. The docking data were in accordance with our experimental results that α-mangostin binds to the substrate binding pocket of ABCG2. In summary, our results indicate that α-mangostin directly and selectively inhibits the transport function of ABCG2 and resensitizes ABCG2-overexpressing MDR cancer cells to chemotherapeutic drugs without altering the expression level of ABCG2. Although studies showing favorable MDR reversal results do not necessarily translate into successful clinical outcomes,¹³ our data suggest that in combination with conventional anticancer agents, α-mangostin could improve oral bioavailability or treatment outcome in cancer patients.

ABBREVIATIONS:

MDR

multidrug resistance

ABC

ATP-binding cassette

FCS

fetal calf serum

PBS

phosphate-buffered saline

CCK-8

Cell Counting Kit-8

IMDM

Iscove’s Modified Dulbecco’s Medium

MTT

3-(4,5-dimethylthiazol-yl)-2,5-diphenyllapatinibrazolium bromide

Vi

sodium orthovanadate

IAAP

iodoarylazidoprazosin